Add to Google Preferred Sources
Corma, a Tel Aviv and San Francisco-based startup, has raised $60 million in seed funding led by Sequoia Capital to build a foundation model specifically for defensive cybersecurity. The company argues that general-purpose AI models are highly capable as attackers but poor as defenders, citing internal simulations where AI attackers succeeded 88% of the time while AI defenders detected only 12% of threats. Corma deploys AI agents that function like virtual employees across an organization’s existing security tools, and reports that early customers have seen threat-response times drop by more than 94% and security coverage expand 15-fold. Founded in 2025 by AI researchers and veterans of Israel’s Unit 8200, the company is already working with Fortune 100 and Fortune 500 firms across healthcare, finance, energy, and critical infrastructure.
Key Elements
A new startup is emerging from stealth with a bold thesis: the same artificial intelligence models that can write code and reason through complex problems are far better at attacking computer networks than defending them. Corma, a company founded in 2025, has raised $60 million in seed funding to flip that script by building a foundation model trained specifically for defensive cybersecurity.
The round was led by Sequoia Capital, with participation from Khosla Ventures and Coatue. The company, which is headquartered in both Tel Aviv and San Francisco, did not disclose its valuation. Corma employs 20 people in its Tel Aviv office and has already deployed its AI agents to Fortune 100 and Fortune 500 companies across healthcare, financial services, energy, critical infrastructure, and retail, according to CEO and co-founder Alon Pluda.
“We train language models for defensive cyber,” Pluda said in an interview with Calcalist. “The models thus become much better at cyber than any other model.”
Rather than selling a conventional software dashboard or alert system, Corma deploys AI agents that operate across an organization’s existing security tools. The system is designed to function like a virtual employee, executing end-to-end defensive tasks without replacing the human workforce. “We don’t replace anyone and we are not a product we provide services,” Pluda explained. “People need to operate cyber products today and need someone to perform the problems and fix them. Companies still have thousands of security people and pay a lot for services. I am not replacing them but making them much more successful and much stronger.”
Corma’s pitch is rooted in a structural imbalance that has emerged in cybersecurity as AI has advanced rapidly. General-purpose models from OpenAI, Anthropic, and Google have become highly proficient at writing and refining code, identifying software bugs, and reasoning through multi-step processes. Those same capabilities map directly to offensive security, enabling attackers to move from using AI as a coding assistant to executing complete, autonomous attack chains. Anthropic’s disclosure of its Mythos AI system demonstrated how far these capabilities have come.
Defensive cybersecurity, however, demands a fundamentally different skill set. Beyond protecting code, security teams must sift through enormous volumes of audit logs, network traffic, and security events, identify faint signals that may only become visible over long periods, and maintain consistent judgment across thousands of decisions. Corma argues that general-purpose models are poorly equipped for this work.
To prove that point, the company ran hundreds of simulations using realistic enterprise environments modeled on Fortune 500 companies, equipped with dozens of security tools. Corma tested leading AI models, including OpenAI’s GPT and Anthropic’s Claude, by first instructing them to act as attackers and plant persistent threats inside the simulated organizations. The same models were then asked to defend the environments and find the threats they had created. The results were stark: the AI attackers succeeded in 88% of the simulations, while the AI defenders detected only 12% of the threats.
“The race to general intelligence in cybersecurity has already begun, and the attackers have a significant head start,” Pluda said. “It requires a complete AI-powered defensive workforce, built from the ground up for cybersecurity, that gives defenders the same speed, sophistication and generalization that AI has already given attackers.”
In early customer deployments, Corma says its system has already demonstrated dramatic results. The company reports that its AI agents reduced threat-response times by more than 94%, expanded security coverage by 15 times across different security functions, and uncovered multi-stage attack campaigns that would otherwise have gone undetected.
| Metric | Result |
|---|---|
| Threat-response time reduction | More than 94% |
| Security coverage expansion | 15x across security functions |
| AI attacker success rate (general-purpose models) | 88% |
| AI defender detection rate (general-purpose models) | 12% |
Note: Data sourced from Corma’s internal simulations and early customer deployments.
Sequoia Capital partner Shaun Maguire framed the investment as a response to the structural speed advantage that agentic AI provides to attackers. “Corma has trained its model for the complexity of real-world attacks and is building the intelligence layer defense actually needs,” Maguire said. “Agentic AI gives attackers a structural speed advantage, but Alon’s hacking talent paired with Corma’s frontier AI research helps companies combat these threats at scale.”
Vinod Khosla, founder of Khosla Ventures, emphasized the broader stakes. “As cyberattacks become more autonomous and scalable, the stakes move beyond data and finances to critical infrastructure, healthcare systems, and essential services where failures can have real-world consequences,” Khosla said. “That is why we need entirely new approaches to cyber defense and teams like Corma pursuing one of the hardest problems in cybersecurity.”
The company’s name is drawn from J.R.R. Tolkien’s The Lord of the Rings. In the Elvish language, “corma” means ring — the all-powerful object at the center of the story. Pluda told Fortune that his company’s product is like that ring, “but this time for the defenders.”
Corma’s founding team blends AI researchers, including alumni of Google and DeepMind, with cybersecurity specialists from Israel’s elite Unit 8200 intelligence corps and major cybersecurity companies. The team has experience spanning foundation-model training, advanced AI systems, and both offensive and defensive cybersecurity operations.
The seed round closed in early 2026, and Pluda indicated that the company is not rushing to raise additional capital or aggressively expand its customer base. “We work with many leading companies and we are not in a hurry to recruit more,” he told Calcalist. The company plans to use the funding to scale its AI models by expanding the data and training that power them, and to grow its team with a focus on hiring top talent across defensive security, AI, and research.
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.
